Maybe we should just list all threads on the first page.
There are tradeoffs. Wise decision makers weigh a variety of considerations.
Exactly a week ago at 6:30PM Eastern time, 109 topics had been posted on within the previous 48 hours. Under the new regime the number of topics is up to …111, which is essentially the same. The data doesn’t always cooperate. Perhaps the increase in threads that we previously observed merely reflected ordinary changes throughout the week. Or maybe board activity has dropped off recently for unexplained reasons.
I ran another experiment. Late Wednesday night, I reposted a thread topic that I had previously asked in February 2006: “What’s the typical kickback percentage?” Alas, in neither case was the question answered. But in 2006 corrupt pol Duke Cunningham (R, CA) had just been convicted of bribery and I had the parenthetical remark “(Duke charged 1%)” in the thread title. Anyhoo, the number of page views was actually higher last week: 140 vs. 126 two years ago: a 9% increase.
These numbers are disappointingly small and probably statistically insignificant.
Back in the 1990s a substantial minority of users had the 640×480 resolution, which has basically disappeared. Today a substantial minority uses 1024 x 768. So the number of pixel rows has increased from 480 to 768 - an increase of 60%. A similar increase in the number of threads would move the page splice from 50 to 80. Blackberry users are another matter.
Science continues. I hope to collect another sample of data tomorrow evening. But frankly methinks the best evidence is impressionistic: do the moderators believe that fewer valid questions are sinking into oblivion? Or has this treatment had no qualitative effect? A comparison of threads that ended on Thursday Aug 20 with those that ended on Thursday Aug 13th might provide an answer. I’ll leave that exercise in judgment to another Doper. ( I pick Thursday, because Friday’s threads still have a chance of resuscitation.)
I could live with 80.
Though I still fret over making changes which may increase the server load, even if slightly, on a system that already often seems overtaxed.
You’re doing good work.
Is there not a extension that could let people choose for themselves? Not that I expect Jerry to be able to get on that any time soon, He seems like a very busy guy,
There might be an extension, but I didn’t come across it on the web. There’s a site devoted to vBulletin hacks, but I suspect that anything like that would have to undergo extensive testing. As it is, this board pushes its software near its design limits so… hacks are not gonna happen soon and that’s probably a good thing.
More data! We’ve popped back up to about 140 posts within the past 2 days, up from about 107 last week, same time. This graph shows the comparison:
http://wm54.inbox.com/thumbs/1a_130ba4_d92543f2_oJ.jpg.thumb
For policy purposes, we can zoom in on the 40 post to 120 post interval:
http://wm54.inbox.com/thumbs/1c_130ba2_4b8b51b6_oJ.jpg.thumb
Note that the 30 hour mark is closer to 113 than 120, so that’s another argument for dialing it back somewhat.
Ok. On Friday, I was thinking that we might try something in the 90-120 range. After viewing Sunday’s surprising data point, I feared that I was looking at a statistical artifact, and that made me lean towards a page splice cutoff of 70-80.
But it seems that Sunday was the outlier, so I might retake Friday’s stance. But we’re working with a small sample size and the data seems a little erratic. And we know for sure that increasing the page splice to a couple bazillion will inconvenience some posters.
Here’s another possibility. Could we set the page splice at 80 and increase and increase the display time to 3 days? Or are the only choices for admins, 1 day, 2 days, 1 week, etc? In essence this might set the default to something like 3 pages of GQ threads, rather than two. Some of us on occasion prowl through the more ripened threads: setting the default to 3 days might encourage this sort of behavior.
This bears repeating. I for one tend to like to defer to the opinions of the board’s fine moderators, but we really should pay due regard to the IT department.
FYI, I used to have the same problem on my iPhone, and the solution is to add a bookmarklet to scroll to the bottom of the page. Not sure if Blackberries support it, but, basically, you create a bookmark whose url is
javascript:scroll(0,document.body.scrollHeight);
and when you select that “bookmark” while on any web page, it takes you to the bottom of the page.
(Before I came across the bookmarklet solution, I just had bookmarks for all the SDMB forums on my iPhone, so I didn’t need to scroll to the “Forum Jump” at the bottom of the page)
I personally prefer the longer listing.
Ok, here’s another snapshot from today, Monday and the previous Wednesday. I compare the three to “Baseline” which is Monday August 17th, a few hours after the beginning of the experiment.
http://wm54.inbox.com/thumbs/1d_130ba1_779efd52_oJ.jpg.thumb
For “after” we have three squiggly lines bunched together, all indicating that more threads are considered for a given time interval beyond nine hours or so. For the baseline we have but a single datapoint, which implies that it’s possible that Monday Aug 17th could be an outlier. But the implication is that changing the page split indeed causes Doper attention to be spread over a wider base of threads. Anyway, I consider this particular exercise complete.
Thanks; I can drop to the bottom of the page easily enough, but that’s not a good solution either. The screenful of boilerplate text at the bottom of every page then causes me to have to scroll up for BB screen after BB screen.
With the tiniest font I get only two threads per screen on my BB. But even on a regular computer screen, scrolling through screen after screen of threads I’ve already seen and rejected is tiresome.
I respect the experimentation that Measure for Measure has been attempting, but the data is thin (one data point for the control group). And as I’ve said before, I don’t think it’s even a proper goal to be rescuing marginal threads for less-frequent visitors.
Moreover, I guess I’m a little taken aback by the way this change was implemented. One poster makes a suggestion that it might be nice to structure the board differently, and major changes result? That’s just so alien to the way things normally seem to get done around here.
There’s been no poll, no announcement, no sticky. Most members probably don’t know about this thread.
There’s been no discussion of server load. Given that changes to the search function have been consistently ruled out on this basis, I’m really surprised that it’s just not an issue here.
I vote again to go back to 50. And even if that’s off the table, I think there should at least be some greater public discussion of this change.
Incidentally, there is a way of meaningfully expanding the sample. If we switch to a page split at 100 threads, then the 150 thread experiment becomes our control.
Tom: Two threads per page: yikes! Is there a find function? (I know, I know: it’s impossible to remember every thread title that you’re interested in.)
I don’t think this constitutes a major change though (except for Blackberry/iPhone users). If it was, this thread would be moving towards a 5 page riot by now.
Anyway, when the Powers that Be weigh the evidence, methinks they should assume that Blackberry users vote for 40 threads per page. Put the page maximalists like myself at something like 115 (down from 120 given the extra data). And most, I can only assume, fall in somewhere between. I still insist though that 5-6 hours of 1st page material in GQ (implied by a page split at 50) is rather short.